README.md

metadata

license: apache-2.0
task_categories:
  - question-answering
  - text-retrieval
language:
  - en
tags:
  - rag
  - narrativeqa
  - reading-comprehension
  - retrieval
size_categories:
  - 10K<n<100K
configs:
  - config_name: answers
    data_files:
      - split: train
        path: answers/train*
      - split: dev
        path: answers/dev*
      - split: test
        path: answers/test*
  - config_name: corpus
    data_files:
      - split: train
        path: corpus/*
  - config_name: qrels
    data_files:
      - split: train
        path: qrels/train*
      - split: dev
        path: qrels/dev*
      - split: test
        path: qrels/test*
  - config_name: queries
    data_files:
      - split: train
        path: queries/train*
      - split: dev
        path: queries/dev*
      - split: test
        path: queries/test*
  - config_name: retrieved_docs
    data_files:
      - split: train
        path: retrieved_docs/train-*
      - split: dev
        path: retrieved_docs/dev-*
      - split: test
        path: retrieved_docs/test-*
dataset_info:
  config_name: retrieved_docs
  features:
    - name: query_id
      dtype: string
    - name: corpus_id
      dtype: string
    - name: rank
      dtype: int64
    - name: retrieval_score
      dtype: float64
    - name: is_relevant
      dtype: bool
  splits:
    - name: train
      num_bytes: 37045044
      num_examples: 327470
    - name: dev
      num_bytes: 3915257
      num_examples: 34610
    - name: test
      num_bytes: 11942607
      num_examples: 105570
  download_size: 6206098
  dataset_size: 52902908

NarrativeQA RAG

Dataset for Retrieval-Augmented Generation (RAG) based on NarrativeQA.

Structure

Subset	Splits	Description
corpus	train (default)	Wikipedia plot summaries shared across all query splits
queries	train, dev, test	Reading comprehension questions
qrels	train, dev, test	Relevance judgments (query ↔ document)
answers	train, dev, test	Reference answers (longest annotated answer)

Dataset statistics

Split	Queries	Corpus
train	32747	1572
dev	3461	1572
test	10557	1572

The corpus is shared across all splits and contains Wikipedia plot summaries (document.summary.text) from the original NarrativeQA documents.

• Dev split: mapped from the original validation split
• Corpus source: unique documents from train, validation and test splits

Source

Component	NarrativeQA resource
Train	train split from deepmind/narrativeqa
Dev	validation split
Test	test split
Corpus	document.summary.text (+ document.summary.title)
Answers	Longest answer text per question

Schema

corpus

{"id": "...", "title": "...", "text": "..."}

queries

{"id": "...", "text": "..."}

qrels

{"query_id": "...", "corpus_id": "...", "score": 1}

answers

{"query_id": "...", "answer": "..."}

Usage

from datasets import load_dataset

corpus = load_dataset("DinoStackAI/narrativeqa-rag", "corpus")["train"]
queries = load_dataset("DinoStackAI/narrativeqa-rag", "queries")
qrels = load_dataset("DinoStackAI/narrativeqa-rag", "qrels")
answers = load_dataset("DinoStackAI/narrativeqa-rag", "answers")

train_queries = queries["train"]
dev_qrels = qrels["dev"]
test_answers = answers["test"]

Citation

NarrativeQA is released under the Apache 2.0 License.

@article{kocisky-etal-2018-narrativeqa,
    title = "The {N}arrative{QA} Reading Comprehension Challenge",
    author = "Ko{\v{c}}isk{\'y}, Tom{\'a}{\v{s}}  and
      Schwarz, Jonathan  and
      Blunsom, Phil  and
      Dyer, Chris  and
      Hermann, Karl Moritz  and
      Melis, G{\'a}bor  and
      Grefenstette, Edward",
    journal = "Transactions of the Association for Computational Linguistics",
    volume = "6",
    year = "2018",
    pages = "317--328",
}